Abstract
I describe the design and iterative implementation of a learning progression for supporting statistical reasoning as students construct data and model chance. From a disciplinary perspective, the learning trajectory is informed by the history of statistics, in which concepts of distribution and variation first arose as accounts of the structure inherent in the variability of measurements. Hence, students were introduced to variability as they repeatedly measured an attribute (most often, length), and then developed statistics as ways of describing ?true? measure and precision. The design of the learning progression was guided by several related principles: (a) posing a series of tasks and situations that students perceived as problematic, thus creating a need for developing mathematical understanding as a means of resolving prospective impasses; (b) creating opportunities for developing representational fluency and meta-representational competence as constituents of conceptual development; (c) introducing statistics as invented measures of the qualities of distribution; and (d) adopting an agentive perspective for orienting student activity, according to which distribution of measures emerged as a result of the collective activity of measurer-agents. Instructional design and assessment design were developed in tandem, so that what we took as evidence for the instructional design was subjected to test as a model of assessment, resulting in revision to each. I conclude with a look at ongoing work to design an assessment system to measure students? understandings of data and statistics, and with some thoughts about prospective synergies between mathematics and science education.
Richard Lehrer